Note: I posted some additional data breakdowns in a comment (7 Nov.)
With the US congressional midterm elections almost upon us, I wanted to put the playing field of 2006 into the context of 1994, the last time the party in control of the House of Representatives swung.
In 1994, fifty six seats held by Democrats swung to the Republican party (four seats swung the other direction). Here is how those seats broke down by Democratic vote in the previous election, 1992. In the following paired lines the first line is a range of votes percentages (of the 2party vote) for the Democratic candidate in 1992, while the second line is the number of such districts, followed by the number of those districts that swung (and the percentage that these swingers represented out of that group).

5051.99
21 11 (52.4)
5254.99
26 11 (42.3)
5559.99
44 18 (40.9)
6064.99
40 10 (25.0)
6569.99
45 4 (8.9)
7079.9
46 2 (4.4)
Interpreting these numbers, we can say that more than half the districts that the Democratic party had won with less than 52% of the twoparty vote in 1992 were seats that swung. (That is somewhat surprisingly low; nearly half of the seats that were won so narrowly in 1992 did not swing, even in a year with a huge national votes swing against their party.) Not surprisingly, as the 1992 margin grows, the share of seats that swung declines. Nonetheless, the Republicans even picked up six seats in which the Democratic incumbent won more than 65% of the vote in 1992 (five of these six were open seats; the incumbent was retiring).
So, what if we compare the 2006 playing field; that is, let’s look at how many seats fit into each of these categories based on their 2004 Republican vote. Again, we have two lines, with the grouping by percentages of the Republican candidate’s vote in the first line and the number of districts in that group on the second line. In addition, I have replicated, in parentheses, the number of districts in this category going into 1994 (for the party then in control, the D’s).

5051.99
6 (21)
5254.99
10 (26)
5559.99
19 (44)
6064.99
38 (40)
6569.99
49 (45)
7070.99
60 (46)
Notice how many fewer close races there were in 2004 compared to 1992. If the Democrats were to have exactly the same “harvest rate” in each group of seats (i.e. 52.4% of the seats the Republican won with under 52%, 42.3% of those won with 52% to 55%, etc.), they would gain about 32 seats. And that is right within the range of current projections–actually, a bit below.
Finally, let’s look at the seats Rothenberg rated late last week as most likely to swing. I have simply taken his categories of seats that are tossups or “tilt” or “lean” one way or the other and indicated the vote from 2004. (In this case, the votes indicated are the Democratic sharae of the twoparty vote; in some cases, my data are missing.)
PURE TOSSUP (20 R, 0 D)
Dem % vote, 2004
* CA 11 (Pombo, R) 39.7
* CT 2 (Simmons, R) —
* CT 4 (Shays, R) 47.6
* FL 16 (Open; Foley, R) 32
* FL 22 (Shaw, R) 36
* IL 6 (Open; Hyde, R) 44.2
* KS 2 (Ryun, R) 42.4
* MN 1 (Gutknecht, R) 37.3
* MN 6 (Open; Kennedy, R) 46.0
* NM 1 (Wilson, R) 45.6
* NY 20 (Sweeney, R) 33.9
* NY 26 (Reynolds, R) 44.4
* OH 1 (Chabot, R) 40.1
* OH 2 (Schmidt, R) 28.3
* PA 4 (Hart, R) 36.3
* PA 6 (Gerlach, R) 49
* PA 8 (Fitzpatrick, R) 43.9
* TX 22 (Open; DeLay, R) 42.7
* VA 2 (Drake, R) 44.9
* WI 8 (Open; Green, R) 29.8
TOSSUP/TILT REPUBLICAN (10 R, 0 D)
* AZ 1 (Renzi, R) 38.2 (48.2 in 2002)
* CA 4 (Doolittle, R) 34.6
* CO 4 (Musgrave, R) 46.7
* ID 1 (Open; Otter, R) 30.5
* KY 3 (Northup, R) 38.5
* KY 4 (Davis, R) 44.7
* NV 3 (Porter, R) 42.6
* NY 25 (Walsh, R) —
* NY 29 (Kuhl, R) 44.6
* WA 8 (Reichert, R) 47.6
TOSSUP/TILT DEMOCRATIC (7 R, 3 D)
* AZ 5 (Hayworth, R) 39.1
* CT 5 (Johnson, R) 39.0
* FL 13 (Open; Harris, R) 44.7
* GA 8 (Marshall, D) 24.4
* GA 12 (Barrow, D) —
* IL 8 (Bean, D) —
* IN 9 (Sodrel, R) 49.7
* NH 2 (Bass, R) 39.6
* NY 24 (Open; Boehlert, R) 37.3
* NC 11 (Taylor, R) 45.1
LEAN REPUBLICAN (3 R, 0 D)
* NJ 7 (Ferguson, R) 42.3
* OH 12 (Tiberi, R) 38
* TX 23 (Bonilla, R) 29.8
LEAN DEMOCRATIC (6 R, 1 D)
* IA 1 (Open; Nussle, R) 44
* IA 3 (Boswell, D) —
* IN 2 (Chocola, R) 45.1
* OH 15 (Pryce, R) 40
* OH 18 (Open; Ney, R) 33.8
* PA 7 (Weldon, R) 40.7
* PA 10 (Sherwood, R) —
That’s a lot of seats in play that were not all that close in 2004, which is good news for Democrats, given how few really close races there were in 2004. By my count, there are eight seats in Rothenberg’s tossups in which the Democratic party won under 40% of the vote in 2004. There are five such seats among the tiltD, and one among the leanD seats. That’s nineteen of the 147 seats the incumbent of the party currently controlling the House won with from 60% to 80% of the twoparty vote in 2004. Seats in that category heading into 1994 swung at a rate of only about 12%. If 12% of the current 147 such seats swung this time, that would be seventeen (of the nineteen such seats Rothenberg identifies as tossup or lean). In other words, while it may seem unlikely that so many seats that looked “safe” after 2004 could really be in jeopardy, the lesson of 1994 is that when there is a strong tide of change, lots of seemingly “safe” seats do indeed fall. It is not only the obvous “marginals” that fall. (Remember, barely half of the seats won by the Democrat in 1992 with under 52% of the vote swung in 1994.)
We will know soon how many of these seats really do swing.